Extremely long hang during live migration at "drive mirror is starting"

[kianb]

When trying to migrate a live vm from a pve1 to pve2 over gigabit ethernet (yes, I have confirmed the speeds)

When trying to migrate a vm with 40gb of storage, it hangs for about 20 minutes as the first image shows at the "drive mirror is starting"

Is this a networking issue? The only thing I notice is that going from pve2 -> pve1 is considetably faster and causes no IO delay on the cpu graph of the pve, but going from pve1->pve2 causes about 12% IO delay on the cpu usage during the migration, is the ssd on pve2 broken? How can I test this?

 

[fiona]

Hi,
do you have discard=on set on the disk? In that case, the LVM-thin target disk will be filled with zeroes first, which is why it can take so long, see: https://forum.proxmox.com/threads/v...-thin-with-discard-results-in-high-i-o.97647/